In this episode of The Connector Podcast, hosts A and B dissect the rapidly evolving global payments ecosystem. They begin by examining how instant payments have forced fraud‑prevention to become real‑time and spotlight New Zealand’s collaborative “Get Verified” utility, a mutualised implementation of the UK’s Confirmation of Pay (COP) that lets competing banks share a trust‑orchestrating platform. The discussion moves to the Legal Entity Identifier (LEI), explaining its role in FATF Recommendation 16 as a universal 20‑digit identifier that reduces counter‑party uncertainty in cross‑border flows. The conversation then shifts to current market frictions, notably the explosive growth of Buy‑Now‑Pay‑Later (BNPL). Citing a 2025 CFPB study showing 63 % of BNPL borrowers hold multiple concurrent loans, the hosts explain how the lack of affordability checks creates systemic risk, and how FICO’s upcoming Score 10 suite will incorporate BNPL data to improve transparency. Next, they explore the importance of local payment preferences for merchants, using Poland’s BLIK mobile‑payment system and the nation’s reliance on parcel lockers as a case study of how mismatched checkout experiences erode conversion. The dialogue then contrasts two digital‑money paradigms: privately issued stablecoins (highlighted by the US Genius Act) and tokenized deposits—bank‑backed digital representations of fiat on permissioned blockchains. The hosts argue that tokenized deposits retain existing regulatory oversight and lower compliance costs, positioning them as the preferred institutional pathway, especially under Europe’s MiCA, DORA, and PSD3 frameworks. Finally, the episode looks ahead to AI‑agent‑driven autonomous payments. India’s live UPI pilot, integrated with AI models like ChatGPT, enables agents to complete purchases without human interaction. Global players such as Mastercard (Agent Pay), Google (AP2), and Visa are developing similar capabilities. The hosts discuss the disruptive threats to traditional banking—erosion of card‑loyalty programs and pressure on net‑interest margins as agents constantly chase optimal rates—and the regulatory headwinds in Europe (PSD2/SCA, GDPR, EU AI Act, liability uncertainty). They conclude with two takeaways: the shift toward instant, automated, and autonomous payment systems is already underway, and financial institutions must build flexible, compliant infrastructure now to stay ahead of the curve.
Chapters
Welcome back to The Deep Dive. Today, we're jumping into something huge, the global payments landscape. It's, well, it's probably the fastest moving part of finance right now, and we're going way past just tapping your phone. We want to look at the deep architecture, how trillions move, and maybe more importantly, what happens when that money starts moving itself. We've got a lot to cover today. We're looking at these new collaborative ways banks are fighting fraud, the whole debate around digital money. And, you know, the different ways regulations are shaping up globally.
That's right. And the mission today really is to help you make sense of it all quickly. We'll unpack these trust blueprints, look at some friction in e-commerce and credit, and definitely contrast stablecoins with tokenized deposits. And then we hit the big one, payments made entirely by AI. It's coming faster than people think.
Okay, let's start right at the beginning. Trust. When payments went instant, that old buffer, you know, the couple of days delay, that just vanished. So fraud prevention had to become instant too. How do you even do that, especially when you've got competing banks involved?
Yeah, that's the key problem. And New Zealand offers a really interesting model here. They looked at the UK's Confirmation of Pay system, or COP. But instead of every bank building its own version, which gets super expensive and complicated, they chose, well, collaboration over just code.
Collaboration over code. Okay, what did that actually look like?
They basically mutualized it. They set up something called Get Verified. It's a central utility owned by the banks themselves. It handles the common stuff, orchestrates the trust between banks, so individual banks didn't need to build massive custom tech stacks. They just plugged into the shared platform.
So, a shared infrastructure for compliance and trust. I get the efficiency, but getting competitors to agree on a central utility? That sounds like it must have taken a lot of wrangling upfront, maybe not cheaper initially.
That's a fair point. The initial setup, getting everyone aligned, probably had its challenges. But the sources suggest the long-term view won out. The speed and future-proofing benefits were huge. And the results seem to bear that out. ANZ Bank reported payment errors dropped by 30% after they implemented COP via this utility. Industry-wide, complaints to the Ombudsman went down 17%. But here's the really clever part, looking ahead. Because they built this shared governance and tech The cost of adding new shared services, like maybe cross-bank identity checks or credit assessments, drops dramatically. The hard part, the trust architecture, is already done. The lesson they learned was, find your Janet. You need someone, a convener, who can actually bridge the gaps between banks, regulators, tech people, make them talk the same language.
Find your Janet. I like that. And building trust like that needs standardized identity, right? Which brings us neatly to the LEI, the Legal Entity Identifier. Why is this 20-digit code suddenly so important globally?
It ties straight into the FATF, the Financial Action Task Force, and their updated travel rule, Recommendation 16. Essentially, it mandates better data sharing on who's sending and receiving money internationally. And the new guidance specifically calls out the LEI as a key piece of data for identifying legal entities in payment messages. They're aiming for full implementation by 2030.
Okay, so why the LEI specifically? What's the value?
Precision and reliability, really, especially for cross-border payments and correspondent banking. Think about it, instead of relying on algorithms trying to match names, which can be super error-prone and subjective, you have a unique, standardized identifier. This helps banks manage risk, particularly sanctions and money laundering risks. They can be much more confident about who their counterparty actually is.
Right, so less guesswork, less risk, potentially more stable global flows because Banks might be less inclined to just pull out of regions deemed too risky due to identity uncertainty.
Exactly. It could help de-risk some of those corridors again.
Okay, that makes sense for the plumbing. Let's shift gears a bit and talk about some of the current friction points, things happening right now in the market, like buy now, pay later. It's everywhere, but regulators seem worried.
They are, and with good reason. BNPL has exploded, but it's creating potential systemic issues. There was a recent CFPB study, I think for 2025. It found that 63% of BNPL borrowers took out multiple loans simultaneously within the year. That's huge overlap.
63%.
Wow.
Wow. That's— that's a lot of stacked loans.
It is. And the core issue is that unlike credit cards, BNPL providers generally haven't been required to check if the borrower can actually afford to repay all these concurrent loans. It's a bit of a regulatory blind spot.
So people could be getting seriously overextended without the system really seeing it clearly.
Precisely. But things are starting to change. FICO, for instance, is updating its main credit scoring model, the FICO Score 10 suite, specifically to include BNPL data. So these previously somewhat hidden loans are now becoming part of the standard credit record. Lenders will get a much clearer picture of someone's total debt load.
Okay, so transparency is catching up. That transparency is also key in e-commerce, right? Especially globally. Why is it so hard for merchants to get good payment acceptance rates across different countries.
Yeah, it's a constant headache for them. And the main reason is often trying to force a one-size-fits-all checkout experience. It just doesn't work. You absolutely have to understand local preferences. Take Poland, for example. Their local mobile payment system, BLIK, is dominant. People trust it, they use it. Trying to push only cards there won't maximize your sales.
It's not just the payment method, though, is it?
No, not at all. It's the whole experience. Again, in Poland, parcel lockers have become incredibly popular, often the preferred delivery method. Why? They're convenient, 247, no missed deliveries. If you're an online merchant ignoring that logistical preference, you're potentially losing customers, leaving money on the table. It's about the entire chain, from click to collection.
The full chain of trust and convenience. Got it. Alright, let's pivot to the money itself— digital assets. We're seeing regulation catch up, like the US Genius Act formally recognizing stablecoins. Does this mean stablecoins are sort of the designated future of digital money?
Well, they definitely grab the headlines. They're privately issued, often run on public blockchains. But from an institutional and regulatory perspective, they carry risks. Is the issuer stable? What's really backing it? And here's something often overlooked: the hidden costs. Getting your actual fiat currency in and out, those on-ramp, off-ramp fees, foreign exchange charges, compliance, can all add up. Sometimes the total cost isn't much better than traditional remittance.
Okay, so if stablecoins have these issues, what's the alternative gaining steam, especially within the regulated banking world?
That would be tokenized deposits. This is a really important distinction. These aren't new private currencies. They are literally existing bank deposits represented digitally, usually on a permissioned or private blockchain. Because it is bank money, just in a tokenized format, it keeps all the existing trust, Regulatory oversight, and accounting clarity. Huh.
But wait, if it's just regular bank money on a private blockchain run by banks, yeah, where's the revolutionary crypto edge? Isn't the point of DLT usually decentralization? Why would banks bother?
That's the crucial point. They're not adopting DLT for the crypto ideology. They're adopting it for efficiency. Tokenized deposits promise lower transaction costs. They align much better with existing compliance frameworks, and they integrate more easily into the bank's current systems compared to, say, integrating various stablecoins. They're regulator-friendly. friendly because they often fall outside the scope of complex new crypto laws like Europe's MiCA or the US Genius Act. For banks, DLT here is a potentially cheaper, faster rail, not a replacement for fiat currency itself.
Okay, so it's DLT for evolution, not revolution. Makes sense.
And this sets up a really interesting global dynamic. You have the US creating frameworks that legitimize stablecoins, possibly boosting the dollar's role in digital flows. Meanwhile, Europe is building this dense regulatory fabric, MiCA, DORA, PSD3, focused on resilience and integrating digital assets safely into the existing financial system, favoring things like tokenized deposits. The big question is whether the euro-based, more integrated rails can scale up fast enough before dollar-denominated stablecoin networks potentially dominate cross-border liquidity. It's a bit of a race.
A race indeed. Okay, we have the regulated present. Now let's look over the horizon at something potentially, well, paradigm shifting. Agenting payments.
Yeah.
We're moving beyond card not present. This is person not present. Explain that. An AI agent is the buyer. Give us that coffee example again.
Right. Imagine you tell your AI assistant, handle my morning coffee. It knows you like a flat white, but its core instruction isn't go to Starbucks, it's get the user a flat white optimally.
Okay.
So, if the queue at your usual place is 15 minutes long today, the AI sees that. It instantly checks nearby cafes, finds one with no queue, orders the flat white there, pays for it using your linked account, and sends the pickup code to your phone, all autonomously. It optimizes for your time and cost, not loyalty to a specific shop.
Whoa, okay, hang on. So my AI could just decide to use a different credit card for every single purchase based on which one offers the best points or lowest interest rate at that exact moment? That feels like it could completely demolish bank loyalty built on, well, inertia.
It's precisely that. It's the, uh, thrilling but also slightly terrifying frontier. And it's not science fiction. It's happening now. India has already launched the world's first national pilot doing exactly this. They've integrated their huge UPI payments network directly into AI models like ChatGPT. BigBasket, the grocery giant, was the first merchant. The whole point is for AI agents to autonomously complete transactions at scale.
India is live with this?
Live pilot, yes. And globally, the big players are all building frameworks. Mastercard has Agent Pay. Google has its AP2 protocol using these signed intent mandates. Visa's working on intelligent commerce. But the sheer scale and unified nature of the India UPI pilot, that's what's really groundbreaking.
Let's talk about the earthquake this sends through traditional finance. You mentioned inertia. What are the biggest threats here?
The agent's only loyalty is to optimization for the user. That just bypasses consumer inertia, which, let's be honest, is a huge asset for banks currently. So threat one, the agent routes every payment to the card or method with the absolute best terms, rewards, rates, whatever. Card loyalty programs get instantly stress tested. Threat two, and this one is maybe deeper structurally, the agent will constantly automatically move your cash balances to wherever the highest yield is offered. Think about bank profits. Around 30% of retail banking profit globally comes from net interest income, the margin between what they pay you on deposits and what they earn lending it out. An AI agent optimizing your savings yield 247 directly directly attacks that core profit engine. It will relentlessly hunt for the best rates.
It basically turns every user into a hyper-rational, high-frequency financial optimizer, whether they consciously want to be or not.
Exactly. But replicating India's scale seems incredibly difficult in places like Europe right now because of the existing regulations, right?
Absolutely. Europe's regulatory framework, while designed for safety, throws up massive roadblocks for this kind of autonomous AI agent. Think about PSD2 and Strong Customer Authentication . That usually requires you, the human, to verify a payment. How does that work when an AI is acting completely on its own? It breaks the model.
Then you have GDPR, the data privacy rules. An AI agent constantly learning your spending habits, making decisions, that potentially clashes with principles like purpose limitation. What exactly is the data being used for? And these agents would almost certainly fall under the high-risk AI category in the new EU AI Act. That triggers a whole host of strict requirements. Transparency, human oversight, record keeping. It's complex.
And the biggest question of all.
Liability. Yes. When the AI messes up, sends money to the wrong place, makes a dodgy transaction, who is legally on the hook? Is it you, the user? You gave the initial instruction.
Yeah.
Is it the AI developer? The bank providing the account? The payment network? The protocol the agent used?
Nobody has a clear answer yet.
Not really. And until that chain of liability is clearly defined by regulators, scaling this kind of truly autonomous payment system in the West is going to be, well, Very challenging.
Okay, so putting this all together, we've covered collaborative trust frameworks, the messy reality of BNPL risk, the stablecoin versus tokenized deposit debate, and this frankly mind-bending future of AI spending. What are the core takeaways for someone trying to navigate all this?
I think there are 2 main things. First, the shift towards instant, automated, and even autonomous systems. It's not hypothetical anymore. It's happening. That India pilot is live right now. The direction of travel is clear. Second, because of that, success and maybe even survival is going to hinge on having robust, flexible, and compliant infrastructure ready to handle these changes safely.
So, don't just react to the latest headline.
Exactly. Don't chase headlines, chase readiness. Financial institutions need to be aligning their technology architecture, their risk models, their product roadmaps now. The goal should be readiness to, you know, Switch these new capabilities on safely in a matter of weeks once the regulatory path is clear, not be stuck in a massive rebuilding project for months or years because you weren't prepared.
Build foundation now to be ready for what's inevitably coming. That makes sense. Infrastructure is key. And think about that AI agent again. We know it's going to optimize everything for the user. Best rate, best points, shortest queue for coffee, as you said. So here's the final thought for you, our listeners, to chew on. What happens to brand loyalty? So much of it is built on habit, maybe inertia, sometimes an emotional connection. What happens to that loyalty when the actual purchase decision is delegated to an impartial, totally rational, calculating AI? Can brand loyalty even survive relentless optimization?
